HumorDB
Abstract
Despite significant advancements in computer vision, understanding complex scenes, particularly those involving humor, remains a substantial challenge. This paper introduces HumorDB, a novel image-only dataset specifically designed to advance visual humor understanding. HumorDB consists of meticulously curated image pairs with contrasting humor ratings, emphasizing subtle visual cues that trigger humor and mitigating potential biases. The dataset enables evaluation through binary classification (Funny or Not Funny), range regression (funniness on a scale from 1 to 10), and pairwise comparison tasks (Which Image is Funnier?), effectively capturing the subjective nature of humor perception. Initial experiments reveal that while vision-only models struggle, vision-language models, particularly those leveraging large language models, show promising results. HumorDB also shows potential as a valuable zero-shot benchmark for powerful large multimodal models.
Type
Publication
Is AI fun? HumorDB: a curated dataset and benchmark to investigate graphical humor
This work introduces HumorDB, a novel dataset designed to advance visual humor understanding in AI systems. The dataset consists of carefully curated image pairs with contrasting humor ratings, emphasizing subtle visual cues that trigger humor while mitigating potential biases.
Key features of HumorDB include:
- Evaluation through multiple tasks: binary classification, range regression, and pairwise comparison
- Focus on capturing the subjective nature of humor perception
- Potential as a zero-shot benchmark for large multimodal models
Our initial experiments reveal that while vision-only models struggle with humor detection, vision-language models, particularly those leveraging large language models, show promising results. This work contributes to pushing the boundaries of AI’s ability to comprehend nuanced human communication, specifically in the domain of visual humor.
